Rotation, scaling and translation invariant object recognition in satellite images

Yıl 2016, Cilt: 58 Sayı: 2, 28 - 49, 01.10.2016

Yusuf Soyman Alparslan Ilgın Hakkı

https://doi.org/10.1501/commua1-2_0000000095

Öz

ROTATION, SCALING AND TRANSLATION INVARIANT OBJECT RECOGNITION IN SATELLITE IMAGES

Öz

In satellite imagery, different shots may be taken according to viewing angle, rotation and scale. Algorithms used to recognize objects in satellite images should find them without being affected from these variations. In this paper, rotation, scaling and translation invariant object recognition in satellite imagery is performed. The criterions affecting the performance of such a system are investigated in a strictly controlled environment. As a result of experimental studies with the usage of different parameters, successful performances have been attained. In consequence of contrasting studies the effects of these parameters upon object recognition have been emphasized

Anahtar Kelimeler

Object recognition, RST invariance, feature extraction, classification, clustering, Bag of Visual Words (BOVW), rigid body, satellite imagery